Epimerisation

Extensive structural analysis of patellamide D and lissoclinamides 4, 5 and 6, confirmed that the amino acids directly N-terminal to the thiazole/thiazoline rings are epimerised to d-stereocentres[61]. Additionally many of the related cyanobactins that contain heterocycles have also been reported to contain d-amino acids, including trunkamide A[62,63], lissoclinamide 7[45,61] and the microcyclamides[64]. For the teneucyclamide and trichamide families of cyanobactins, the presence of d-amino acids has been proposed, but has not been confirmed in all products[65,66]. Since ribosomes are selective for l- amino acids, epimerisation must be a PTM[45]_.

From our current understanding of the biosynthetic pathway, we cannot state definitively when, during patellamide maturation, epimerisation takes place. Epimerisation can be in theory either acid or base catalysed. While the↵-carbon adjacent to the thiazoline is not particularly acidic, perhaps favouring acid catalysis over base catalysis, thiazolines are chemically unstable under acidic conditions and readily ring open to give cysteine[50]. Assuming then a base catalysed mechanism, epimerisation would seem favourable after thiazoline formation, and prior to its oxidation to a thiazole, thus avoiding the need to disrupt aromaticity to stabilise the build up of negative charge that occurs following proton abstraction. Therefore the following base catalysed mechanism, where proton abstraction is stabilised through conjugation onto the imine nitrogen seems viable[45] (Fig. 1.13).

Figure 1.13: Epimerisation mechanismA base abstracts a proton at the↵-carbon adjacent to the thiazoline ring. This is stabilsed via conjugation and subsequent protonation of the imine nitrogen. Re-protonation occurs at the ↵-carbon on the

opposite face.

Since it is not explicit whether oxidation directly succeeds heterocyclisation, follows leader sequence removal, or only occurs on the cyclic peptide, the precise timing of

Chapter 1. Introduction 23

epimerisation is uncertain. Furthermore it is unclear as to whether epimerisation is a spontaneous or an enzymatic process in cyanobactin biosynthesis. Early work by Wipf

et al. demonstrated that if lissoclinamide 7 was synthesised with l-valine adjacent to the thiazoline, it would epimerise tod-valine following treatment with pyridine[45] _(Fig. 1.14), consistant with the ↵-carbon adjacent to a thiazoline being susceptable to non- enzymatic epimerisation.

Figure 1.14: Epimerisation of lissoclinamide 7 Epimerisation of unnatural l- isomer to the naturald-isomer using 5-10 equivalents of pyridine.

However, the required 5-10 equivalents of pyridine and temperatures of 60°C to a↵ord the correct isomer[45] do not reflect the physiological conditions of the cyanobacteria, and so it cannot be considered proof of spontaneous epimerisation in nature. More convincing was the observation by Salvatellaet al. that an enantiomer of trunkamide A synthesied withl-phenylalanine, when solubilised in a combination of 30 % d6-DMSO/70 % CDCl3for a series of NMR experiments, epimerised to the naturald-enantiomer within days[67]_{(Fig. 1.15). Analyses of solution structures of the isomers revealed trunkamide A} undergoes a significant conformational rearrangment following inversion of the F residue, resulting in a less sterically hindered, planar structure and allows for two, new hydrogen bonding interactions[67] (Fig. 1.15).

It was proposed that this increase in conformational stability of the macrocycle is the driving force for the epimerisation reaction[67]. This agrees with epimerisation being a spontaneous process, and that if it is, it most likely occurs on the macrocycle and not on the linear peptide. It is difficult to rationalise a similar thermodynamic driving force for a spontaneous epimerisation on the flexible linear peptide. Such a

Chapter 1. Introduction 24

Figure 1.15: Epimerisation of trunkamide A Unnaturall-isomer spontaneously epimerises to naturald-isomer within days. (a) 2D stick representation; (b) 3D solution

NMR structures determined by Salvatellaet al. (figure taken from paper)[67]

‘conformational stability’ hypothesis might be sufficient to explain the conservation of d-stereocentres throughout the cyanobactin superfamily[21] _{and why the natural} compounds are isolated as single enantiomers and not diastereoisomers[45]. That said, cyanobactins are structurally highly diverse, with known compounds of di↵erent ring sizes and di↵erent constitutent amino acids. Therefore, it might not be that every cyanobactin will experience the same conformational pressures favouring epimerisation, as observed for trunkamide A. d-Stereocentres have so far always been reported adjacent to thiazoline/thiazole heterocycles, and never next to oxazoline/oxazole heterocycles, even though these residues are also potentially labile to epimerisation. Such selectivity, whether chemical or regioselectivity, might be under control of an enzyme catalyst. While there is evidence supporting a spontaneous epimerisation of synthetic cyanobactins, the requirement of an epimerase in their biosynthesis has not been explicitely ruled out. However no epimerase activity has been identified in any of the biosynthetic enzymes, nor is epimerase activity predicted based on sequence homology of

Chapter 1. Introduction 25

the remaining uncharacterised gene products in the biosynthetic cluster[27]. Epimerase, racemase and isomerase enzymes are common place throughout biology, exhibiting a vast diversity in structure and size, and subtle variation in their catalytic mechanisms, especially concerning the amino acid combination utilised as acids and bases[68–72]_{. This} makes it difficult to predict epimerase activity amongst the uncharacterised proteins in cyanobactin pathways. If epimerisation is catalysed by one of these uncharacterised gene products, this would potentially represent a novel class of epimerase.